Structure and composition of single Pt–Ru electrocatalyst ... · different components:...

37
https://cimav.repositorioinstitucional.mx/ 1 Structure and composition of single Pt–Ru electrocatalyst nanoparticles supported on multiwall carbon nanotubes Francisco Paraguay-Delgado, Marek Malac and Gabriel Alonso-Núñez Abstract Individual Pt-Ru nanoparticles (NPs) supported on multiwall carbon nanotubes (MWCNTs) synthesized by microemulsion method were characterized by nano beam diffraction (NBD) and high resolution imaging in transmission electron microscopy (TEM). Comparing the TEM images and NBD to simulations provided insight into particle composition, structure and morphology in three dimensions. In particular, the NBD allowed us to detect various components of the individual NPs that would be difficult to observe otherwise. We find that the NPs contain four different components: Pt–RuO2,Pt–Ru, RuO2 and metallic Pt. Often an individual NP is composed of more than one component. The most frequently encountered external morphology is close to a spherical shape and ~3.7 nm in diameter. The collective properties of NPs’ assemblies were studied by thermogravimetry, differential thermal analysis and x-ray diffraction. The results allowed us to gain some insight into the relation of the NPs’ structure and composition with their catalytic performance, and revealed the presence of components not detectable by bulk methods. The electrocatalytic properties were evaluated by CO stripping, methanol oxidation and oxygen reduction. Bulk characterization methods miss many properties and structures present in the sample due to low volume fraction and due to overlap of reflections. Single NPs should be analyzed to obtain reliable indication of sample composition.

Transcript of Structure and composition of single Pt–Ru electrocatalyst ... · different components:...

  • https://cimav.repositorioinstitucional.mx/

    1

    Structure and composition of single Pt–Ru electrocatalyst nanoparticles supported on multiwall carbon nanotubes

    Francisco Paraguay-Delgado, Marek Malac and Gabriel Alonso-Núñez

    Abstract

    Individual Pt-Ru nanoparticles (NPs) supported on multiwall carbon

    nanotubes (MWCNTs) synthesized by microemulsion method were characterized by

    nano beam diffraction (NBD) and high resolution imaging in transmission electron

    microscopy (TEM). Comparing the TEM images and NBD to simulations provided

    insight into particle composition, structure and morphology in three dimensions. In

    particular, the NBD allowed us to detect various components of the individual NPs

    that would be difficult to observe otherwise. We find that the NPs contain four

    different components: Pt–RuO2,Pt–Ru, RuO2 and metallic Pt. Often an individual

    NP is composed of more than one component. The most frequently encountered

    external morphology is close to a spherical shape and ~3.7 nm in diameter. The

    collective properties of NPs’ assemblies were studied by thermogravimetry,

    differential thermal analysis and x-ray diffraction. The results allowed us to gain

    some insight into the relation of the NPs’ structure and composition with their

    catalytic performance, and revealed the presence of components not detectable by

    bulk methods. The electrocatalytic properties were evaluated by CO stripping,

    methanol oxidation and oxygen reduction. Bulk characterization methods miss many

    properties and structures present in the sample due to low volume fraction and due

    to overlap of reflections. Single NPs should be analyzed to obtain reliable indication

    of sample composition.

  • https://cimav.repositorioinstitucional.mx/

    2

    Keywords: single Pt-Ru nanoparticle structure, electrocatalyst, nano-beam

    electron diffraction, microemulsion

    1. Introduction The performance of catalyst nanoparticles can be strongly affected by the

    presence of a small amount of highly active component in the nanoparticle

    population, differences in size, composition, shape and surface sites [1, 2].

    Understanding the structure–properties relation is therefore key in identifying the

    active component of the NP population and in selecting a suitable synthesis method.

    Various structures have been proposed for Pt–Ru NPs on multiwall carbon

    nanotubes (MWCNTs) and black carbon, including core–shell NPs and Pt–Ru alloys

    [3], but experimental evidence supporting a particular case is limited. MWCNTs are

    used as an electro-catalyst support primarily due to their one-dimensional

    morphology and good thermal and electrical conductivity [4]. The MWCNT support

    performance often exceeds that of black carbon support [2, 5, 6]. The orientation of

    the individual Pt–Ru NPs relative to the MWCNT support also affects catalytic

    selectivity and sensitivity [7], but it is difficult to study. Here, we report extensive

    investigation about Pt–Ru NPs structure deposited on MWCNT and their

    electrocatalytic performance. In particular, we show that nano beam diffraction

    (NBD) has the capability to reveal the structure of nanoparticles made of several

    components.

    A transmission electron microscope (TEM) allows one to image individual

    NPs, but interpretation of unknown structure from high resolution TEM (HRTEM)

    images alone is notoriously difficult [8]. Electron beam-induced change of orientation

  • https://cimav.repositorioinstitucional.mx/

    3

    and/or structure [9]isof critical importance when studying small metallic particles.

    Nano beam diffraction (NBD) in a TEM mode using parallel illumination combined

    with low dose (low magnification) survey imaging allows one to obtain structural

    information at low irradiation dose [10]. NBD allows detection of small (about 5%)

    volume fractions of phases present within individual NPs by appearance of

    characteristic reflection for each phase [11]. NBD patterns interpretation is similar to

    standard selected area electron diffraction (SAED) and tends to be more reliable

    than interpretation of HRTEM images revealing internal structure of individual NPs.

    The difficulty in HRTEM interpretation arises from the phase contrast nature of

    HRTEM imaging where the contrast can reverse with sample thickness even for very

    small (a monolayer) sample thickness [12]. Annular dark field image, acquired by

    scanning transmission electron microscopy (STEM) images are easier to interpret,

    but often require high irradiation dose to collect high resolution images resulting in

    concerns on electron irradiation damage of the samples [10].

    Platinum–ruthenium (Pt–Ru) bimetallic NPs can be used as an anode catalyst

    for direct methanol fuel cells (DMFC) [13] and direct ethanol fuel cells (DEFC) [14,

    15]. The interest in bimetallic catalyst NPs arises from enhanced catalytic

    performance compared to elemental NPs. For example, binary Au–Pd nanoparticles

    [16] have been used for solvent-free oxidation of primary carbon–hydrogen bonds in

    toluene. The interest in Pt–Ru stems from their promise to improve DMFC and

    DEFC performance and understand the mechanism. In particular, when Pt is used

    together with Ru, the NPs show high tolerance to presence of carbon monoxide

    (CO), which is frequently present in a fuel cell. Many methods can be utilized for

  • https://cimav.repositorioinstitucional.mx/

    4

    synthesis of bimetallic nanoparticles. They include co-precipitation, hydrothermal

    and inverse microemul-sion methods [17]. A suitable synthesis method must

    produce NPs with high catalytic activity and selectivity and with long term stability

    [18, 19]. Synthesis often yields samples that contain many types of NPs with wide

    variations in performance that depend on individual structure and morphology [20].

    The NBD investigations appear to provide an avenue to gain insights needed to

    establish the relations between NP structure and properties.

    2. Experimental section

    2.1. Sample synthesis

    Pt–Ru NPs on MWCNT were prepared in a two-step process, first MWCNTs

    were prepared. Second, Pt–Ru NPs were synthesized and deposited onto

    MWCNTs. The MWCNTs were synthesized by spray pyrolysis method using iron

    ferrocene (Fe (C5H5)2) as a catalyst. The ferrocene was dissolved in toluene as

    described by Núñez et al [21]. The synthesis leaves Fe, Fe carbide or Fe oxides

    particles both on the outside and inside of the as-prepared MWCNTs. The Fe

    particles on the outside of the MWCNTs were removed completely by nitric acid

    (HNO3) wash, but the iron compound particles inside the MWCNTs could not be

  • https://cimav.repositorioinstitucional.mx/

    5

    removed, because they were encapsulated by the MWCNTs. The cleaned MWCNTs

    were then rinsed several times with tri-distilled water until the MWCNTs suspension

    reached pH = 7 and they were then filtered and dried. The Pt–Ru NPs were then

    dispersed onto MWCNTs.

    To obtain bimetallic Pt–Ru particles deposited onto MWCNTs, the Pt–Ru

    nanoparticles were first synthesized by standard microemulsion technique [22] using

    the water–oil system. The Brij 30 compound was used as a surfactant and heptane

    was used as the oil. The metallic precursor for Pt and Ru were hexacloroplatinate

    (H2PtCl6) and ruthenium chloride (RuCl3), respectively. The reducing agent was

    sodium boron hydride (NaBH4) for both precursors. The microemulsion molar ratio

    between water and surfactant was 3:18. To ensure that the Pt–Ru nanoparticles

    were bimetalic, the Ru precursor microemulsion was added into the Pt

    microemulsion, resulting in bimetallic intermicelar microemulsion. This bimetallic

    microemul-sion was then added to MWCNTs suspension in heptane and it was

    stirred for 24 h for adequate distribution and deposition of all intermicellar Pt–Ru

    microemulsion on the MWCNTs surface.

    Two batches of samples were prepared: one with 25 wt% nanoparticle

    loading and the other with 10 wt% loading. The expected fraction of bimetallic NPs

    with respect to MWCNTs was estimated from the synthesis parameters. For the

    nominal 25 wt% batch, the expected fractions of bimetallic NPs with respect to

    MWCNTs were 25 wt% (Pt 19.7 wt%and Ru 5.6 wt%, respectively, see table 1). The

    nominal 10 wt% nanoparticle loading sample was prepared with the expected

    loading Pt 7.6 wt% and Ru 1.96 wt%. The two batches of samples at 25 w% and 10

  • https://cimav.repositorioinstitucional.mx/

    6

    wt% loading allowed us to compare the electrochemical results. For easy

    identification the samples were named according to their composition, as shown in

    table 1, together with the exact composition obtained by inverse microemulsion

    method.

    2.2. Characterization and electrochemical activity measurement

    Thermogravimetry (TG) and differential thermal analysis (DTA) technique for

    bulk samples were performed using TA-instruments SDT 2920. Samples were

    heated in a platinum-cup from room temperature up to 750 °C with 10 °C min−1

    heating rate. The TG-DTA studies were performed under dry air flow to oxidize the

    MWCNTs, removing carbon and leaving the metallic compound only. The main

    purpose of analysis by TG-DTA is to quantify the metallic (Pt–Ru) content in the

    functionalized MWCNTs and to obtain a baseline reference value. X-ray diffraction

    (XRD) patterns were obtained using an analytical X-pert pro alpha-1 diffractometer

    using Cu Kα radiation (λ = 0.15406 nm) in the θ − 2θ Bragg–Brentano arrangement.

    Nano beam electron diffraction (NBD) patterns were acquired using a Hitachi

    HF-3300 transmission electron microscope (TEM), equipped with cold field emission

    gun and three-lens condenser system, operated at 300 kV. The instrument permits

    the formation of a sub-5 nm probe with beam convergence angle below 0.3 mrad [8,

    9]. Under such conditions, it was possible obtain parallel-beam-like diffraction

    patterns from individual Pt–Ru NPs on MWCNTs. High angle annular dark field

    (HAADF) images were acquired on probe aberration-corrected JEOL JEM2200FS

    scanning transmission electron microscope operated at 200 kV. The beam

    convergence semi-angle was approximately 28 mrad. The HAADF collection semi-

  • https://cimav.repositorioinstitucional.mx/

    7

    angle was from 100 mrad to 170 mrad. The microscope is also equipped with an

    Oxford Inca model x-ray energy dispersive spectroscopy (EDS) for elemental

    microanalysis. TEM specimens were prepared by dispersing and sonicating the

    sample in methanol for 2 min, a drop of this suspension was then put onto a holey

    carbon coated copper grid and allowed to dry in laboratory air. The electron

    diffraction patterns were simulated using CaRine crystallography V3.1 and

    SimulaTEM [23] software. The SimulaTEM allows one to take into account the

    shape and size of nanoparticle in diffraction pattern simulations. The images and

    diffraction patterns were simulated using an appropriate unit cell for internal structure

    of a nanoparticle and Mackay icosahedra to account for particle shape [24]. The

    atomic coordinates within the MacKay icosahedra were generated using an existing

    Matlab™ code [25].

    The electrocatalytic properties were tested by dynamic potential

    electrochemistry. The measurements were performed at 25 °C using a potentiostat

    in a thermostated three-electrode electrochemistry cell. A plate of glassy carbon

    (GC) was used as a working electrode (diameter 3 mm). A reversible hydrogen

    electrode (RHE) was connected to the working electrode compartment through a

    Luggin capillary in 0.5 M H2SO4 aqueous electrolyte, which was the reference

    electrode. The oxygen reduction reaction (ORR) measurements were carried out

    using rotating disk electrode (RDE) with 0.07 cm2 geometric surface area, at

    different angular speed rotation. The same sample was used in cyclic

    voltammogram (CV) and OOR study techniques. To evaluate suitability of the NP-

    loaded MWCNTs as an electrode the samples were prepared by mixing 10 mg of the

  • https://cimav.repositorioinstitucional.mx/

    8

    NP-loaded MWCNTs in 0.25 mL Nafion® (5 wt% in water/aliphatic alcohol solution)

    with 1.25 mL ultra-pure water. The samples were sonicated for 1 h; 3 μL of the

    resulting solution was deposited onto the GC electrode and was dried under nitrogen

    atmosphere. CVs were recorded in nitrogen-saturated electrolyte solution between

    0.05 and 1.4 V at 50 mV s−1 as a scan rate. First, 20 cycles were performed to

    stabilize the system. Linear current–potential curves for ORR were recorded from

    1.0 to 0.05 V versus RHE in oxygen-saturated solution at different angular scan

    speed rates. CVs were also recorded in nitrogen-saturated methanol 0.5 M in

    H2SO4 0.5 M with the same starting and stopping potentials. The ORR curves were

    also measured in oxygen-saturated and methanol at 0.5 M H2SO4 + 0.5 M

    CH3OH; these measurements were made using rotating electrode at different

    angular scan speed rates, in order to evaluate both the tolerance to pollutants during

    the ORR and to know their selectivity of the material.

    3. Results and discussion

    3.1. Average properties from large volume sample

  • https://cimav.repositorioinstitucional.mx/

    9

    Bulk sample analysis can provide an estimate of sample average composition

    and structure. However, an attempt to link sample (catalytic) properties with

    composition and structure obtained by bulk analytical methods can result in a

    misleading interpretation: a particular structure or composition that can be below

    detection limit of bulk analytical techniques can have a profound effect on catalytic

    performance of the sample [26]. Due to their widespread use and to obtain reference

    values, we first evaluate the sample bulk properties, composition and structure by

    TGA-DTA, XRD and broad electron beam selected area diffraction in a TEM.

    Thermogravimetric measurements show the thermal stability of the

    functionalized MWCNTs. They were used to quantify the metallic content of the Pt–

    Ru NPs-loaded on MWCNTs. Purified MWCNTs without Pt–Ru loading were used

    as a reference. Figure 1 shows TG-DTA curves for the purified MWCNTs (red) and

    PtRu25 (blue) samples, respectively. The red curve shows that the purified

    MWCNTs start to decompose at 530 °C and the decomposition appears to be

    complete at 650 °C. The reaction is exothermic, as shown by the DTA (red dots). In

    the case of the sample loaded with Pt–Ru NPs, there is a broad region of mass loss

    from 120 to 350 °C likely resulting from the removal of organic residue from

    surfactant used to synthesize the NPs. The decomposition and oxidation of the

    PtRu25 sample is also an exothermic reaction, (blue dotted DTA curve). The

    decomposition starts at 350 and ends at 550 °C. This is about 180 °C lower than for

    the MWCNTs reference sample without NPs. A likely explanation is that the active

    Pt–Ru NPs on the surface of MWCNTs catalyze carbon oxidation, suggesting that

    the amount of the metal on the MWCNTs surface has a considerable influence on

  • https://cimav.repositorioinstitucional.mx/

    10

    MWCNTs thermal stability. A possible mechanism is that MWCTNs oxidation

    accelerates the desorption of CO2 from carboxyl groups of the functionalized

    MWCNTs surface. At 550 °C the MWCNTs with NPs are completely burned

    leaving about 26% of the weight of PtRu25 sample corresponding to the Pt–Ru

    residue. For comparison, the purified MWCNTs weight residue is about 3.3% arising

    from encapsulated iron catalyst used for the synthesis of the MWCNTs. The

    difference in the residual weight of the Pt–Ru-loaded and pure MWCNTs samples

    provides an estimate of the Pt–Ru fraction contained in the functionalized PtRu25

    sample. The fraction of metallic residue in the PtRu25 sample is close to the nominal

    metallic load quantity used in the synthesis process (25 wt% see table 1). The

    metallic residue quantity (3.3 wt%) observed in the purified MWCNTs is attributed to

    the iron encapsulated inside of the MWCNTs during the synthesis process that

    cannot be removed by the HNO3 wash step described earlier.

  • https://cimav.repositorioinstitucional.mx/

    11

    X-ray diffraction was performed to further investigate the bulk properties of the

    NP-loaded MWCNTs sample. This technique provides good statistics suitable for

    investigation of average structure of the bulk material, but cannot detect phases

    present in less than about 10% volume. Figure 2 shows XRD patterns obtained from

    three samples: purified MWCNTs, Pt10 and PtRu25. All of them exhibit two strong

    diffraction peaks with 3.4 Å and 1.74 Å, d-spacings belonging to (002) and (004)

    arising from MWCNTs graphitic planes, respectively. Two additional peaks are

    present in the PtRu25 sample, observed at 2.26 Å and 1.96 Å. They belong to (111)

    and (200) metallic Pt d-spacings with fcc structure. Using XRD the structure of the

    sample with Pt–Ru nanoparticles can be attributed solely to fcc Pt nanoparticles; the

    Ru, Pt–Ru and other phases cannot be reliably detected. The diffraction peaks

    around 44 degrees arise from Fe carbides or metallic Fe which were encapsulated

    inside the MWCNTs, during the synthesis process.

  • https://cimav.repositorioinstitucional.mx/

    12

    Comparing the pure MWCNTs and Pt–Ru loaded sample XRD data indicates

    that functionalization of the MWCNT by Pt–Ru NPs does not result in modification of

    the MWCNTs structure. Broad beam selected area electron diffraction (SAED)

    patterns obtained in a TEM (not shown) collected from PtRu25 sample contains

    rings which can be assigned solely to fcc Pt NPs and to MWCNT.

    TG-DTA, XRD and SAED techniques provide average structure of the

    samples, but cannot be used to study composition and structure of individual NPs.

    Average properties provide good statistics as the data is averaged over a large

    number of NPs, but suffers from poor ability to detect low volume fraction of

    particular structure that can be present and can be important for the catalyst

    performance. Unfortunately, often average properties obtained from bulk samples

    are the only evidence used. Various structures for Pt–Ru NPs, such as core–shell

    [27], alloys [28, 29], linked monometallic nanoparticles [18], bimetallic [30] and

    others [31] were proposed, but there is no clear experimental evidence for any

    particular one. Therefore, we investigate the structure and composition of individual

    Pt–Ru NPs in the following paragraph.

    3.2. Characterization of single Pt–Ru nanoparticles

    For practical applications of NPs, such as DMFC and DEFC, it is important to

    understand the role of the structure and composition variations among particles [32–

    34]. The first step in linking the structure and morphology and catalytic activity of

    NPs to their structure is to obtain a structure of a sufficient number of individual NPs.

    This will provide an insight into the variation in compositions and the structure of the

    NPs. In this paragraph we discuss the elemental composition and structural

  • https://cimav.repositorioinstitucional.mx/

    13

    characterization of individual NPs.

    Quantitative elemental analysis of individual NPs was performed by EDS in a

    probe-corrected electron microscope JEOL 2200 FS. The probe was positioned on

    individual NPs in scanning transmission electron microscopy (STEM) mode. The

    EDS spectra show that Pt and Ru ratio varies for each particle. Figure 3 shows the

    elemental composition values distribution from approximately sixty particles. The

    measured Ru content is between zero and 34 wt% while the Pt content is between

    62 wt% and 100 wt%. On average, the particles contained about 14 wt% of Ru and

    86 wt% of Pt, respectively.

  • https://cimav.repositorioinstitucional.mx/

    14

    The 14 wt% Ru content is lower than the 28 wt% nominal content expected

    from the NPs synthesis parameters (see table 1). Figure 3 shows that NPs

    composition varies from 100 wt% Pt to particles with nearly 35 wt% of Ru. It is not

    expected that the Pt/Ru ratio would be affected by electron beam induced radiation

    damage [35] although the NP structure and oxygen content can be affected.

    The Pt–Ru NPs size distribution and spatial separation on the MWCNTs will

    be discussed here for the PtRu25 sample. The NPs spatial distribution on MWCNTs

    as well as their size distribution was obtained by bright field TEM and STEM in bright

    field and high angle annular dark field (HAADF) modes. Figures 4 and 5 show typical

    morphology of Pt–Ru NPs and their location on the outside surface of MWCNTs.

    The NPs were always localized outside of the MWCNTS when the sample was tilted

    up to ±50 degrees.

    Most of the Pt–Ru NPs are well separated and have approximately uniform

    distribution along the MWCNTs. We did not observe significant agglomeration of the

    NPs. The presence of lattice fringes in figure 5(b) indicates that individual NPs are

    crystalline. The most frequently observed d-spacings coresponds to (111) planes of

    Pt. The NPs exhibit round projected shape, with weak facetting. Although the

  • https://cimav.repositorioinstitucional.mx/

    15

    composition measured by EDS (see figure 3) indicates variation of the Pt to Ru ratio

    in individual NPs, the lattice spacing and shape seen in figures 4 and 5(b) do not

    allow differentiation of the various types of NPs in composition: all have

    approximately round shape and exhibit plane spacing close to (111) planes of fcc Pt.

    Figure 6 shows the histogram of particle size distribution (PSD). The PSD was

    based on data from both BFTEM and HAADF STEM images, although HAADF is

    more convenient due to its high contrast [36]. The average particle diameter is D =

    (3.7 ± 0.4) nm. The distribution appears to be log-normal. The small standard

    deviation, 0.4 nm, indicates narrow particle size distribution. The 3.7 nm mean

    particle diameter corresponds to approximately 1500 atoms within each individual

    particle, i.e. about 380 unit cells within each nanoparticle. This narrow distribution

    demonstrates the methodology synthesis; it is good for controlling the size of NPs.

    It is notoriously difficult to interpret unknown structure from an HRTEM or

    STEM images alone. Furthermore, the electron irradiation dose needed to obtain a

    high resolution TEM image tends to be high, possibly leading to sample radiation

    damage for example by removal of the light elements (oxygen) and consequent

    change of NP structure [33]. NBD provides interpretable information at doses

    several orders of magnitude lower than needed for HRTEM images [8]. Using NBD

    can also detect phases present in small volume fractions (down to about 5%) in

    individual NPs rather than in the entire bulk sample. NBD also allows one to analyze

    NPs with orientations away from low-index zone axis that are not suitable for

    HRTEM or STEM analysis [9]. In figures 7 to 10 we present example NBD patterns

    collected from individual NPs; these were compared with simulated NBD patterns.

  • https://cimav.repositorioinstitucional.mx/

    16

    Figure 7(a) shows HRTEM image of a single particle on an MWCNT. The

    0.34 nm spacing of graphite sheets within the MWCNT walls can be clearly seen

    and used as an in situ calibration standard for camera length and for images, they

    are marked with two black arrows in the image.

    Figure 7(b) shows experimental NBD pattern from the same nanoparticle as

    in figure 7(a). The two symmetric reflections marked with arrows close to the central

    beam belong to the 0.34 nm d-spacing of MWCNT graphitic sheets. All other

    reflections originate from the Pt–Ru nanoparticle. The pattern indexing of figure 7(b)

    is shown in figure 7(c). All observed d-spacings arise are from RuO2 with tetragonal

    structure with the [12-3] direction close to parallel with the incident electron beam.

    There is a disagreement between expected angle involving for (111) and (2–10)

  • https://cimav.repositorioinstitucional.mx/

    17

    planes (should be approximately 77.2 degrees) while the observed angle is

    approximately 82 degrees, but the measured d distance for hkl spacing suggests

    RuO2 as the only possibility. There are more additional spots on the first and second

    circle marked in figure 7(c). The disagreement between measured and expected

    angle values for bulk RuO2 can be explained by deformation arising from the small

    size of the RuO2 particle compared to bulk. Similar observations were reported in

    gold and nickel [37]. Figure 7(d) shows simulated NBD for tetragonal RuO2 along

    the [12-3] zone axis. Figures 7(b) and (d) show a good agreement between

    experiment and simulation for RuO2. Figure 7 and its interpretation shows that NBD

    allows one to identify a single crystallographic phase that was not interpretable from

    HRTEM image alone.

    In figure 8 we show that NBD allows one to detect and unscramble

    composition of an individual NP made up of two phases. Figure 8(a) shows two

    isolated particles on an MWCNT. The corresponding NBD pattern from one of them,

    marked by a circle (figure 8(a)) is shown in figure 8(b). The two spots marked with

    the arrows near the center belong to the 0.34 nm planes of the MWCNT. The NBD

    reflections in the figure 8(b) can be assigned to two crystallographic phases: one set

    of reflections belongs to metallic Pt while the other belongs to tetragonal RuO2

    phase. Figure 8(c) shows the metallic fcc Pt reciprocal lattice oriented with the [1-1-

    2] direction parallel to the incident beam. Figure 8(d) shows the tetragonal RuO2

    with [-1-23] axis parallel to the electron beam. Using the known structure for both fcc

    Pt and tetragonal RuO2 it is possible to simulate the diffraction pattern for each

    particle to confirm the indexing, as shown in figures 8(e) and (f), respectively. Based

  • https://cimav.repositorioinstitucional.mx/

    18

    on the NBD pattern in figure 8(b) this nanoparticle is composed of fcc Pt and

    tetragonal RuO2. Both particles have parallel planes, which are marked by circles in

    figures 8(c) and (d) revealing that the Pt (402) planes are parallel to RuO2 (−242)

    planes. Therefore, the Pt {111} (0.265 nm) planes interface with the RuO2 {200}

    (0.251 nm) planes. The measured angles, in addition to the d-spacing, were the key

    to detection of both fcc Pt and RuO2 in this NP.

    Figure 9(a) shows bright field TEM image of particles on another MWCNT.

    The particle orientation is such that d-spacing cannot be seen in an HRTEM image.

    Yet the NBD in figure 9(b), obtained from NP marked by a red circle in figure 9(a)

    shows interpretable reflections. Reflections from the MWCNT are along the green

    line in figure 9(b) showing (002) and (004) planes, respectively. The rings shown in

    figure 9(b) belong to metallic Ru (red) and Pt (blue) as labeled in figure 9(d). The

    strong reflections of metallic fcc Pt are along the [-123] zone axis, as shown in figure

    9(c). Therefore, this particle is composed of metallic fcc Pt and metallic hexagonal

    Ru; it is a bimetallic particle.

    The analysis of these typical NPs interpretation demonstrated how the

    information that is not available from bulk analytical methods can be obtained when

    individual NPs are investigated by NBD. The information on composition and phases

    present in individual particles can be crucial for understanding the catalytic

    performance of the samples.

    In this paragraph we demonstrate how shape, in addition to structure, of an

    NP can be inferred from NBD patterns. Figure 10(a) shows an HRTEM of

    approximately round-shaped particle on an MWCNT. The corresponding NBD from

  • https://cimav.repositorioinstitucional.mx/

    19

    this particle is shown in figure 10(b). Most reflections in the NBD are curved and

    elongated rather than spots. The appearance of these diffraction spots arises from

    shape of the nanoparticles [10] and can therefore be used to obtain information on

    the NP shape. The reflections of the first d-spacing circle appear to be doubled while

    on the third circle the spots are elongated and curved. The relative intensity and

    shape of the reflections have mirror symmetry with respect to a line indicated by the

    red arrow in figure 10(b). Figure 10(c) shows that all reflections are on a circle with

    d-spacings values that belong to metallic fcc Pt particle. Figure 10(d) shows five-fold

    mirror symmetry of the pattern as observed in quasicrystals [38].

    To investigate the origin of the curved spots and their symmetry, we simulated

    a diffraction pattern for a nanoparticle with Mackay icosahedra shape. The simulated

    particle was built using fcc Pt atom coordinates generated using a Matlab script [23].

    The simulated particle was composed of 1500 Pt atoms as expected for the

    measured ∼3.7 nm NP diameter. Diffraction patterns for this particle shape were

    simulated using SimulaTEM software [21].

  • https://cimav.repositorioinstitucional.mx/

    20

  • https://cimav.repositorioinstitucional.mx/

    21

    Figure 11(a) shows DP oriented exactly along the [011] zone axis. This

    simulated pattern shows clear five-fold symmetry similar to the experimental pattern

    in figure 10(b). The Mackay icosahedra overall shape is shown in the top left inset of

    figures 11(a) to (d). Figure 11(a) simulates a pattern exhibiting ten spots on the first

    circle with elongated, curved shape. This can be clearly observed in the central part

    of the simulated DPs; shown in the inset at the bottom right of figures 11(a) to (d).

    The simulated patterns shown in figures 11(b), (c) and (d) were generated by

    rotating the simulated particle around its x-axis 4°, 8° and −10° tilt angles,

    respectively, remaining close to the expected [011] zone axis. The diffraction

    patterns retain five-fold symmetry, but in the first, the d-spacing circle shows two

    spots close to one another while the curved shape of the spots remains unchanged.

    The simulated patterns for 4° and 10°tilt angle, show a horizontal mirror symmetry

  • https://cimav.repositorioinstitucional.mx/

    22

    that is not observed when the particle is oriented exactly on zone axis indicating that

    exact orientation can be retrieved from the NBD patterns of NPs. This can be of

    importance when studying orientation of NPs relative to a supporting substrate. The

    simulated pattern which appears to a best match to the experimental data in figure

    10(b) is shown in figure 11(f), for this purpose the simulated NP oriented along [011]

    zone axis was rotated around x-axis and y-axis by 5° and 3° tilting angle,

    respectively.

    Consequently, the NP shown in figure 10(a) is likely to be a Mackay

    icosahedra Pt particle; similar shaped NPs were reported for Au, but not for Pt

    although both Pt and Au have the same fcc structure [39].

    The NBD results show that, although of similar morphology and size as

  • https://cimav.repositorioinstitucional.mx/

    23

    observed in TEM and STEM images, the Pt–Ru nanoparticles are composed of

    several distinct structures and compositions such as RuO2,Pt–RuO2,Pt–Ru and fcc

    Pt particles. The HREM imaging does not allow one to distinguish the phases and

    could be hindered by radiation damage even in the highest resolution TEM/STEM

    instruments available; NBD on the other hand provided enough information for

    reliable identification of structure even in the case where two compounds are

    present in each NP and allowed to retrieve the shape of an NP.

    The fact that the Pt–Ru NPs are made up of various structures and

    compounds can have significant impact on understanding the DMFC and DEFC

    electro catalysis, because the catalytic action of the various types (in terms of

    composition and structure) NPs is likely to vary.

  • https://cimav.repositorioinstitucional.mx/

    24

  • https://cimav.repositorioinstitucional.mx/

    25

    For example, it was reported that the RuO2 linked to Pt resulting in the same

    NP improves the electocatalytic oxidation of methanol for DMFC [31]. It was also

    reported that an optimum Pt and Ru composition is desirable for improved DMFC

    and DEFC performance [32].

    3.3. Electrochemical performance

    The paragraphs above discuss the shape, elemental composition, crystal

    structure and size distribution of the Pt/Ru NPs. Although their size distribution is

    very narrow, the EDS mapping revealed that the chemical composition of the NPs is

    far from uniform and the NBD revealed that a single NP can be composed of

    multiple components including oxides of Pt and Ru. In this section we discuss the

    electrochemical properties of the PtRu25 material including carbon monoxide (CO)

    stripping voltammetry, methanol oxidation activity and oxygen reaction activity.

    We then attempt to make a link between the performance and the

    composition. We also include the electrocatalytic results from samples Pt10 and

    PtRu10 for comparison purpose with PtRu25. We focus on the performance of the

    PtRu25 sample which was studied in detail by EDS and NBD techniques.

    3.3.1. Carbon monoxide stripping voltammetry.

    One of the main challenges to overcome in the fuel cell technology is

    developing at the anode electrocatalysts that are tolerant to CO at ~50 ppm

    concentration. It is well known that CO strongly adsorbs on the Pt active sites

    preventing oxidation of H2 at those sites. Carbon monoxide stripping voltametry can

    be used to determine the electrochemically active surface (EAS) and the number of

    active sites of the catalyst. This technique is used to study the electrocatalytic

  • https://cimav.repositorioinstitucional.mx/

    26

    reaction and oxidation from CO to CO2 on the surface of the nanoparticles [40].

    Figure 12 shows the cyclic voltammograms (CVs) for CO stripping test obtained

    from Pt10, PtRu10 and PtRu25 samples. For the last sample the increasing voltage

    of the CV curve shows the following processes: CO adsorption, oxidation from CO to

    CO2 and CO2 molecules desorption-peak located around 0.67 V. From the

    characteristic oxidation peak intensities, corresponding to CO oxidation to CO2, the

    Pt–Ru compounds are higher than for the pure Pt NPs on MWCNTs (sample Pt10).

    Additionally, the peak shifts toward lower voltage with increasing Ru content

    (changes from 0.83 V to 0.67 V for Pt10 and PtRu25 samples, respectively).

    The peak voltage shifts toward lower values for up to at least 25 wt% Pt-Ru

    on MWCNT. Therefore, this shift can be attributed to the increased Ru content, the

    NBD reveals a variety of Ru metallic and oxides linked to Pt NPs, thus making it

    difficult to assign the voltage shift due to the particular compound. In the PtRu25

    material, the Ru presence facilitates the CO oxidation due to the OH species

    chemisorbed on the Ru active sites, following reactions 1 and 2 [41].

  • https://cimav.repositorioinstitucional.mx/

    27

    Here QCO is the charge arising from CO oxidation, E is the potential (V), ν

    the sweep rate (5 mV s−1) and j the current density (A cm−2). The integration limit,

    i.e. the oxidation range from CO to CO2, was taken over the range marked by blue

    arrows for PtRu25 sample in figure 12. The measurements of EAS were determined

    by dividing QCO value by 195 uC cm−2 (charge density of bulk polycrystalline Pt

    value). The resulting EAS values are shown in table 2.

    Table 2 shows that the highest EAS value was observed for the PtRu25

    sample. The EAS values are obtained by assuming CO adsorption on the Pt side

    part of the NPs only. No absorption is assumed on the, Ru or RuO2 part of the NPs.

    This is in agreement with equation (2) which shows that the CO adsorption takes

    place only on the Pt surface and not on the surface of Ru or RuO2 [38]. The surface

    area of exposed Pt atoms is reduced due to the presence of Ru compounds on the

    surface of NPs. However, the number of active Pt surface sites on the compound

    particle increases because the Ru and its compounds linked to Pt protect or avoid

    the poisoning of the Pt EAS ‘sites’. The mechanism is still being discussed [42].

    While our results do not provide understanding of the mechanism, the NBD results

    reveal and confirm the existence of the various Ru and Ru oxides which can be

    critical in shedding some light on it.

  • https://cimav.repositorioinstitucional.mx/

    28

    Figure 12 shows that the adsorption (negative j values) and desorption

    (positive j values) parts of the CV curve are symmetric for PtRu25 sample. It

    indicates that the reaction has 100%yields, it is complete; no CO remains unreacted

    in the PtRu25 sample. The Pt10 and PtRu10 samples do not exhibit such symmetry

    of CV curves. The difference between PtRu25 and Pt10 and PtRu10 samples

    suggests, there is an optimum composition for Pt–Ru NPs.

    Figures 4 and 5 show that the NPs are well dispersed on the MWCNTs

    suggesting that the EAS cannot be increased further by improving NPs dispersion.

    Consequently the EAS limit could be set by the amount of available Ru and its

    oxides linked to Pt. The determined EAS values and high tolerance to CO contents

    in the sample PtRu25 suggest that the material may be suitable for use as an anode

    for PEM fuel cells.

    3.3.2. Methanol oxidation activity.

    This type of experiment provides a measure of a material's ability to oxidize

    methanol to determine if the material can be used as an anode. The experiment was

    carried out at potential rate of change 50 mV s−1 in 0.5 M H2SO4 + 0.5 M CH3OH

  • https://cimav.repositorioinstitucional.mx/

    29

    electrolyte with nitrogen gas purges. The CH3OH electro-oxidation follows two

    alternative reaction pathways depending on whether CO adsorbed on the

    electrocatalyst material.

    Figure 13 shows CVs for samples’ suitability for methanol oxidation. The y-

    axis value for the Pt10 sample is on the right hand side while for PtRu10 and

    PtRu25, the y-axis values are on the left hand side. Figure 13 shows that the

    electrochemical activity of Pt–Ru on MWCNTs increases with increasing Pt–Ru

    loading up to the 25 wt%, the maximum loading investigated in this work. The

    forward scan for the PtRu25 exhibits an onset potential Eonset = 0.51 V for

    methanol oxidation. The current density increases rapidly with potential and reaches

    a maximum current density jmax = 8.1 mA cm−2 at Emax = 0.84 V. The peak at 0.84

    V can be attributed to the oxidation of methanol on catalyst surfaces. During the

    reverse scan, the peak current density reaches 6.03 mA cm−2 at 0.62 V, which can

    be ascribed to the incomplete removal of carbonaceous species originating from

    partially oxidized methanol formed in the forward scan [43].

  • https://cimav.repositorioinstitucional.mx/

    30

    The Eonset and Emax values for samples Pt10 and PtRu10 are shown in table 3.

    The Pt10 sample has the lowest jmax value. The increase in Ru content leads to

    corresponding increase in jmax. The jmax value for the PtRu25 sample increased 60

    times (compared to the Pt10 sample). At the same time, the onset and maximum

    peak potentials of methanol oxidation are observed at 0.56 and 0.86 V for PtRu10,

    respectively, which is more positive than the corresponding value of Pt10 sample.

    Furthermore the PtRu25 sample has higher tolerance to CO poisoning compared to

    only Pt10, as witnessed by the higher amplitude of the forward and reverse peaks in

    the CV for PtRu25 sample. It appears that the ability of the PtRu25 material to

    tolerate CO makes a possible candidate for PEM fuel cell application.

    3.3.3. Oxygen reduction reaction. Oxygen reduction reaction (ORR) measures

    the capacity of the material to generate oxygen ions by reacting with H+ to form

    water. The ORR curves were acquired at several angular velocities of a rotating

    electrode (400, 900, 1225, 1600, and 2500 rpm) for the PtRu25 sample in the

    presence of molecular oxygen in 0.5 M H2SO4 at 25 °C with 5 mV s−1 voltage scan

  • https://cimav.repositorioinstitucional.mx/

    31

    rate.

    Figure 14 shows ORR curves for the PtRu25 sample. First, to verify whether

    the oxygen reduction occurs in the sample, a control experiment was performed in

    N2 atmosphere. No current flow was observed in this experiment at 1600 rpm.

    Typically, three distinct regions of the ORR curves can be observed as a function of

    the RHE voltage (RHE is a reference electrode—see section 1.2).

    Within region I, the current is independent from the rotation speed. In region II

    the electrochemical reaction is controlled by both kinetic and diffusion processes. In

    region III the mass transfer diffusion current depends on the rotational speed. All

    these three regions are indicated in the lower part of the figure 14.

    The polarization voltage curves for different angular velocities in oxygen

    atmosphere change show systematic increase in current density j with increasing

    angular velocity. This can perhaps be assigned to enhanced diffusion of oxygen

    towards the active Pt–Ru on MWCNT material. In the PtRu25 sample current flow

    starts at about 0.88 V bias voltage, while for the Pt10 sample the onset is at 0.75 V.

    The maximum current density at 2500 rpm due to oxygen reduction at 0.3 V for Pt10

  • https://cimav.repositorioinstitucional.mx/

    32

    sample appears to be about −2.75 mA cm−2, while the PtRu25 sample exhibits a

    higher current density of −4.9 mA cm−2 under the same conditions. Figure 14 also

    shows the result for purified MWCNT; it starts to oxidize around 0.2 V cathode bias.

    Not surprisingly, the purified MWCNTs do not exhibit catalytic activity for oxygen

    reduction. The results in figure 14 indicate that the sample PtRu25 may be suitable

    for ORR and generates acceptable cathode current density.

    Figure 15 illustrates the methanol tolerance of the samples. The ORR

    experiment was repeated maintaining the same experimental conditions as in figure

    14 but with methanol added. The measurement was carried out using 0.5 M H2SO4

    + 0.5 M CH3OH solution saturated with oxygen and nitrogen. The PtRu25 sample

    under nitrogen atmosphere (i.e. without oxygen) shows strong oxidation current

    density peak at 0.77 V. Experiments under oxygen saturation do not show this peak

    that is the characteristic response of methanol oxidation. This means that PtRu25

    sample shows selectivity to oxygen reduction in the presence of methanol. A control

    experiment for MWCNTs without Pt–Ru NPs yielded no measurable current,

    confirming that the MWCNTs are inert without the presence of the NPs.

    The PtRu25 material exhibits a potential 0.86 V versus RHE (figure 15) for

    methanol oxidation. But in comparison the potential for ORR is at 0.96 V versus

    RHE (figure 14). Then, the PtRu25 material may be a good candidate for oxygen

    reduction reaction (despite the methanol oxidation) which is an important

    consideration for DMFC applications. The key is that the PtRu25 selectively reduces

    oxygen even in the presence of fuel (methanol).

  • https://cimav.repositorioinstitucional.mx/

    33

    The other important characteristic of the electrocatalysts is their capacity to

    generate current and the current limit under both experiments ORR and ORR in the

    presence of methanol (figures 14 and 15, respectively). The difference in observed

    current densities at 2500 rpm and 0.3 V between Pt10 and PtRu25 samples may be

    worth noting. The current density in the presence of methanol (fuel) (figure 15) is 3.4

    and 5.7 mA cm−2 for Pt10 and PtRu25 samples, respectively. This difference in

    values indicates that the PtRu25 yields 67% more current than Pt10. Without

    methanol presence (figure 14) the maximum current density for PtRu25 is

    5mAcm−2, which is lower than in the presence of methanol as a consequence of

    oxygen presence.

    4. Conclusions The catalytic performance of a material could depend on trace amount of a

    phase that may not be detectable by bulk methods, such as XRD and selected area

    diffraction. Individual particles must be analyzed to detect phases that are present in

    small amounts. HRTEM or ADF STEM may appear to be a suitable choice, but in

    many cases a high resolution TEM or STEM image is not sufficient to reliably detect

    the presence of various phases in NPs and determine their structure and mutual

    orientation. The HRTEM imaging limitations arise from phase contrast nature of the

    image contrast; correct interpretation of contrast of phase images is far from trivial

    even when radiation damage is not a concern. Annular dark field STEM methods

    may offer images easier to interpret, but typically require high electron irradiation

    dose resulting in damage of the sample. Both HRTEM and ADF STEM methods

    need the studied NP to be close to a low index zone axis. In our example XRD,

  • https://cimav.repositorioinstitucional.mx/

    34

    SAED and thermogravimetry techniques show average properties and incorrectly

    identify all the particles such as metallic Pt particles. EDS revealed that the

    elemental composition of individual particles spanned many compositions from pure

    Ru to pure Pt. Nano beam diffraction (NBD) of individual NPs allows one to directly

    interpret the structure and orientation of phases within individual NPs. Using NBD it

    was possible to identify the presence of RuO2,Pt–RuO2,Pt–Ru and Pt particles on

    MWCNTs. An example of Pt nanoparticle was given where the shape (faceted

    Macky icosahedra) was retrieved from NBD pattern. While interpretation of such

    experiments is tedious, they have the potential to provide insight into the structure–

    properties relation of catalyst nanoparticles. This knowledge in turn can be used to

    tune the NP synthesis process to optimize the material catalytic properties. From the

    electrocatalytic test, the studied material exhibits acceptable performance to be

    considered as a cathode or anode according to the particular interest application.

    Acknowledgments The authors acknowledge the National Institute for Nanotechnology,

    Edmonton, Canada for the access to Hitachi HF-3300 TEM and to National

    Nanotechnology Lab in CIMAV Chihuahua Mexico for the access to the probe

    corrected JEOL 2200 FS. We thank C Ornelas for kindly providing technical support

    at CIMAV. The electrochemistry results were discussed intensely by Rosa Maria

    Felix and Yadira Gochi at Institute Tecnologico de Tijuana and Oaxaca, Mexico,

    respectively. We thank the Laboratoire de Catalyse en Chimie Organique, Equipe

    Electrocatalyse, Université de Poitiers France for the experimental electrochemistry

    results. We thank Dr N Vante for enabling the stay of our student Aldo Gago. The

  • https://cimav.repositorioinstitucional.mx/

    35

    advice on HF-3300 optics by Dr Y Taniguchi (Hitachi High Technologies Corp) is

    greatly appreciated. We would also like to thank projects CONACYT 155388,

    174689, DGAPA-PAPIIT: IN104714-3.

    References [1] Sambur J B and Chen P 2014 Annu. Rev. Phys. Chem. 65 395–422

    [2] Shan X et al 2012 Nat. Nanotechnol. 7 668–72

    [3] Alayoglu S, Nilekar A U, Mavrikakis M and Eichhorn B 2008 Nat.

    Mater. 7 333–8

    [4] De Volder M F L, Tawfick S H, Baughman R H and Hart A J 2013

    Science 339 535–9

    [5] Jeng K T, Chien C C, Hsu N Y, Yen S C, Chiou S D, Lin S H and

    Huang W M 2006 J. Power Sources 16097–104

    [6] Hull R V, Li L, Xing Y and Chusuei C C 2006 Chem. Mater. 18 1780–8

    [7] Lee I, Morales R, Albiter M A and Zaera F 2008 PNAS 105 15241–6

    [8] José-Yacamán M, Marın-Almazo M and Ascencio J A 2001 J. Mol.

    Catal. A 173 61–74

    [9] Wu M Y, Smeets R M M, Zandbergen M, Ziese U, Krapf D, Batson P E,

    Dekker N H, Dekker C and Zandbergen H W 2009 Nano Lett. 9 479–84

    [10] Malac M, Beleggia M, Taniguchi Y, Egerton R F and Zhu Y 2008

    Ultramicroscopy 109 14–21

    [11] Malac M, Wang F, Egerton R and Taniguchi Y 2007 Microsc.

    Microanal. 13 558–9

    [12] Reimer L and Kohl H 2008 Springer Series in Optical Science vol 36

  • https://cimav.repositorioinstitucional.mx/

    36

    (Berlin: Springer)

    [13] Nitani H, Honda Y, Horioka R, Ohara K, Kawaguchi T, Nakagawa T,

    Seino S, Yamamoto T A and Daimon H 2007 Mater. Res. Soc. Symp. Proc. 1056

    HH03–H25

    [14] Şen S, Şen F and Gökağaç G 2011 Phys. Chem. Chem. Phys. 13

    6784–92

    [15] Switzer E, Atanassov P and Datye A 2010 US Patent 7,670,988,B2

    [16] Kesavan L et al 2011 Science 331 195–9

    [17] Shukla A and Neubert R H H 2005 J. Phys. 65 1097–108

    [18] Zhang L, Kim J, Chen H M, Nan F, Dudeck K, Liu R S, Botton G A and

    Zhang J 2011 J. Power Sources 196 9117–23

    [19] Yamamoto T A, Kageyama S, Seino S, Nitsni H, Nakagawa T, Horioka R,

    Honda Y, Ueno K and Daimon H2006 Appl. Catal. A 396 68–75

    [20] Yin Y 2011 Nano Res. 4 1–2

    [21] Alonso-Nuñez G, Lara-Romero J, Paraguay-Delgado F, Sánchez-

    Castañeda F M and Jiménez-Sandoval S2010 J. Exp. Nanosci. 5 52–60

    [22] Eriksson S, Neylen U, Rojas S and Boutonnet M 2004 Applied

    Catalysis A: General 265 207–19

    [23] Gomez-Rodriguez A, Beltran-del-Rio L M and Herrera-Becerra R 2010

    Ultramicroscopy 110 95–104

    [24] Kuo K H 2002 Struct. Chem. 13 221–30

    [25] Dmitry T 2010 Diploma Thesis University of Duisburg-Essen, Germany

    (www.uni-due.de/agfarle/PDF/dipl-doc/DiplTyutyunnikov.pdf)

  • https://cimav.repositorioinstitucional.mx/

    37

    [26] Sasaki K, Wang J X, Balasubramanian M, McBreen J, Uribe F and Adzic

    R R 2004 Electrochim. Acta 493873–7

    [27] Goto S, Li Y, Nagasawa N, Senoo T, Noda K, Kudo Y, Maesaka A and

    Hatazawa T 2008 Mater. Res. Soc. Symp. Proc. 1127 T07–01

    [28] Li L and Xing Y 2007 J. Phys. Chem. C 111 2803–8

    [29] Zhang X and Chan K Y 2003 Chem. Mater. 15 451–9

    [30] Ueji M, Harada M and Kimura Y 2008 J. Colloid Interface Sci. 322 358

    [31] Jha N, Reddy A L M, Shaijumon M M, Rajalakshmi N and Ramaprabhu

    S 2008 Int. J. Hydrog. Energy 33427–33

    [32] Hiromi C, Inoue M, Taguchi A and Abe T 2011 Electrochim. Acta 56

    8438–45

    [33] Peng F, Zhou Ch., Wang H, Yu H, Liang J and Yang J 2009 Catal.

    Comun. 10 533–7

    Structure and composition of single Pt–Ru electrocatalyst nanoparticles supported on multiwall carbon nanotubes1. Introduction2. Experimental section2.1. Sample synthesis2.2. Characterization and electrochemical activity measurement

    3. Results and discussion3.1. Average properties from large volume sample3.2. Characterization of single Pt–Ru nanoparticles3.3. Electrochemical performance3.3.1. Carbon monoxide stripping voltammetry.3.3.2. Methanol oxidation activity.

    4. ConclusionsAcknowledgmentsReferences